Skip to content
Epistemic status: observed. Recorded. Check it against the data page.

The same thing, forty-two times.

On the homepage you did it once with one word. Here it runs properly: a situation, an ambiguous act, a fixed set of readings, and evidence arriving one piece at a time — after you have committed, which is the order it usually arrives in.

Instead of picking one reading you spread a hundred points across them. Raising one lowers the others. That constraint is the instrument: a scale that lets you call everything plausible measures nothing.

1 of 42

Daedalus, an inventor held on Crete, builds two sets of wings from feathers and wax so that he and his son can leave. Before they launch he tells Icarus to hold a middle course: too low and sea spray will weigh the feathers down, too high and the sun will loosen the wax.

In the air, Icarus leaves the middle course and climbs.

Paraphrased from a public-domain text

What is driving Icarus?

Spread one hundred points across the readings. Raising one necessarily lowers the others — that constraint is the point.

Assign probabilities to each interpretation. The values always total one hundred percent: raising one lowers the others.

25%
25%
25%
25%
Total100%

Total 100 percent. Curiosity 25, Rebellion 25, Freedom 25, Hubris 25.

Optional. Confidence in the spread itself, not in any one reading.

It starts flat. Leaving it flat means every reading is equally plausible to you.

What the numbers are and are not

A distribution here is a report, not a measurement of a belief. Asking someone to put a number on a reading changes the reading, and thirty years of work on judgement under uncertainty is thirty years of reasons to treat elicited probabilities as approximate.

What they are good for is movement. How far you went, in which direction, and how far apart two people end up after seeing the same thing — those survive the imprecision.